The Jeju Agricultural Research and Extension Services has introduced an artificial intelligence (AI) and 3D modeling system to improve forecasting for the island's citrus harvests.
Yield predictions for Jeju's mandarin oranges, one of South Korea's key agricultural exports, have traditionally relied on field surveys. Inspectors visited 320 sample orchards three times per year to count blossoms, measure fruit size and sugar levels, and estimate yields. This approach required intensive labor, was vulnerable to manpower shortages during peak summer heat, and was affected by early harvesting.
The new system integrates 10 years of field survey records with microclimate data from the Korea Meteorological Administration, along with vegetation indices generated from satellites and drones. Using multiple machine-learning models, researchers can automatically predict yields based on blossom-to-leaf ratios, fruit counts and sizes, planting density, and elevation.
During the mid-season survey in August, researchers tested the system by photographing citrus trees with smartphones. The uploaded images generated 3D models of trees within 40 minutes, allowing AI to calculate fruit numbers and dimensions.
"AI-driven forecasting reduces labor intensity, minimizes human error, and raises accuracy," said Kim Tae-woo, head of the Agricultural Digital Center. "By strengthening predictive reliability through digital transformation, we aim to enhance the competitiveness of Jeju's citrus industry."
The rollout is part of South Korea's wider agricultural digitalization program. Jeju's citrus sector is being used as a pilot for implementing advanced forecasting technologies to reduce inconsistencies and improve production planning across the supply chain.
Source: The Korea Bizwire